10482
Biochemistry 2002, 41, 10482-10489
Guest-Host Crosstalk in an Angiogenin-RNase A Chimeric Protein†,‡ Daniel E. Holloway,§ Robert Shapiro,| Michelle C. Hares,§ Demetres D. Leonidas,§,⊥ and K. Ravi Acharya*,§ Department of Biology and Biochemistry, South Building, UniVersity of Bath, ClaVerton Down, Bath BA2 7AY, United Kingdom, and Center for Biochemical and Biophysical Sciences and Medicine and Department of Pathology, HarVard Medical School, Boston, Massachusetts 02115 ReceiVed May 17, 2002; ReVised Manuscript ReceiVed June 25, 2002
ABSTRACT: Angiogenin and ribonuclease A share 33% sequence identity but have distinct functions. Angiogenin is a potent inducer of angiogenesis that is only weakly ribonucleolytic, whereas ribonuclease A is a robust ribonuclease that is not angiogenic. A chimera (“ARH-I”), in which angiogenin residues 58-70 are replaced with residues 59-73 of ribonuclease A, has intermediate ribonucleolytic potency and no angiogenic activity. Here we report a crystal structure of ARH-I that reveals the molecular basis for these characteristics. The ribonuclease A-derived (guest) segment adopts a structure largely similar to that in ribonuclease A, and successfully converts this region from a cell-binding site to a purine-binding site. At the same time, its presence causes complex changes in the angiogenin-derived (host) portion that account for much of the increased ribonuclease activity of ARH-I. Guest-host interactions of this type probably occur more generally in protein chimeras, emphasizing the importance of direct structural information for understanding the functional behavior of such molecules.
The proteomes of complex organisms are comprised largely of families and superfamilies. Most, if not all, of the individual members of these groups serve distinct roles, with variations ranging from subtle to extremely dramatic. One approach for studying the molecular basis for this functional diversity has been to generate chimeric proteins in which residues or segments from one homologue are transplanted into another (1, 2). In principle, the properties of hybrids can provide considerable insight into the functional individuality of the two components. Bovine pancreatic ribonuclease (RNase A,1 EC 3.1.27.5) and human angiogenin (Ang) provide a particularly intriguing system with which to investigate structure-function relationships among homologues. RNase A is a digestive enzyme with high catalytic efficiency toward general RNA substrates (3, 4). Ang was discovered as a tumor-derived angiogenesis factor and found unexpectedly to be a member of the pancreatic RNase superfamily (5-7). Although Ang contains the full catalytic apparatus of RNase A, its ribonucleolytic potency is some 4-7 orders of magnitude lower (1, 8). This weak activity is not a mere vestige of evolution: mutational studies have demonstrated that Ang must have an intact enzymatic active site to induce angiogenesis (9, 10). None† This work was supported by Cancer Research UK Grant SP2354/ 102 (to K.R.A.), Medical Research Council UK Grant 9540039 (to K.R.A.), and National Institutes of Health Grant CA-88738 (to R.S.). ‡ The atomic coordinates have been deposited in the Protein Data Bank (entry 1GV7). * To whom correspondence should be addressed. Telephone: +441225-386238. Fax: +44-1225-386779. E-mail:
[email protected]. § University of Bath. | Harvard Medical School. ⊥ Present address: Institute of Biological Research and Biotechnology, The National Hellenic Research Foundation, 48 Vas. Constantinou Ave., Athens 11635, Greece. 1 Abbreviations: Ang, angiogenin; ARH-I, angiogenin-ribonuclease A hybrid I; RNase A, bovine pancreatic ribonuclease.
theless, RNase A and all other superfamily members that have been tested are devoid of angiogenic activity. Comparison of the sequences of Ang and RNase A reveals that one functionally important segment of RNase A, residues 63-74, is poorly conserved in Ang (Figure 1a): the intrasegment disulfide bridge (65/72) is absent, there is only one apparent sequence identity, and Ang contains two fewer residues overall. Early crystallographic studies (3, 4, 11) had shown that the RNase segment between Cys 65 and Cys 72 forms one face of the B2 purine-binding subsite, and might influence the catalytic site as well. To evaluate the significance of the dissimilarity between Ang and RNase A in this region, Harper and Vallee (1) prepared a hybrid protein (ARH-I) in which residues 58-70 of Ang were replaced with the corresponding residues, 59-73, of RNase A (Figure 1a). The regional substitution enhanced enzymatic activity by up to several hundred-fold, conferred a more RNase A-like specificity at B2, and increased the chemical reactivity of catalytic residues. These changes suggested that the RNase A guest segment in ARH-I is well-accommodated into the Ang host and acts much like it does in RNase A. ARH-I was also found to have greatly reduced angiogenic activity (1), reflecting the involvement of the substituted segment in the cellular interactions of Ang (7, 12). In a complementary study, Raines et al. reported that the reverse substitution (in which residues 59-73 of RNase A are replaced with residues 58-70 of Ang) confers angiogenic activity upon RNase A and considerably lowers its ribonucleolytic activity (13). The crystal structure of Ang (14) confirmed that the threedimensional (3D) structure of residues 58-70 differs markedly from that of its counterpart in RNase A, but also indicated that the impact of the RNase A-derived segment in ARH-I might be more complex than first envisioned. Specifically, it revealed a potential functional connection in ARH-I, between the substituted segment and the B1 pyri-
10.1021/bi026151r CCC: $22.00 © 2002 American Chemical Society Published on Web 07/24/2002
Crystal Structure of Angiogenin-RNase A Chimera
Biochemistry, Vol. 41, No. 33, 2002 10483 Table 1: X-ray Data Collection and Refinement Statistics data collection statistics resolution (Å) no. of measured reflections no. of unique reflections completenessa (%) Rmergea,b (%) I/σ(I)a refinement statistics resolution (Å) Rcrystc (%) Rfreed (%) no. of reflections used no. of protein atoms no. of citrate atoms no. of water atoms deviations from ideality (rms) bond lengths (Å) bond angles (deg) dihedrals (deg) impropers (deg) average B-factor (Å2) main chain atoms side chain atoms all protein atoms citrate atoms water atoms
FIGURE 1: Structural overview. (a) Sequences of Ang (green), RNase A (red), and ARH-I (colored according to protein of origin) in the region of mutagenesis. The alignment is that given in ref 1 and does not take the 3D structures into consideration. Residues that appear to be conserved are highlighted, and disulfide-bonded cysteines are bridged. (b) Schematic representation of ARH-I highlighting strands Β1-Β7 and helices H1-H4. The Ang-derived component is green; the RNase A-derived component is red, and disulfide bridges are yellow. (c) Superposition of the R-carbon traces of ARH-I (gold), Ang (1B1I; green), and RNase A (7RSA; red). The two C-terminal residues of ARH-I are not visible in the crystal structure.
midine-binding site, that does not exist in either parent protein. Further kinetic and mutational studies on ARH-I demonstrated this linkage, while leaving its precise nature unresolved (15). Here we report a crystal structure of ARH-I at 2.1 Å resolution that provides the details of the interactions between the RNase A-derived and Ang-derived regions, and accounts for the characteristic catalytic activity of the hybrid protein. EXPERIMENTAL PROCEDURES Protein Purification, Crystallization, and Diffraction Measurements. ARH-I was prepared from the soluble extract of recombinant Escherichia coli W3110 by cation-exchange chromatography on SP-Sepharose and Mono-S media as described previously (1, 16), and further purified by C4 reversed-phase HPLC (17). The protein was crystallized by the hanging drop vapor diffusion method. A 4 µL drop (pH 5.8) containing 10 mg/mL protein, 7.5% (w/v) PEG 4000, 4% (v/v) tert-butyl alcohol, 0.15 M ammonium acetate, and 0.05 M sodium citrate was equilibrated at 16 °C against a reservoir (0.8 mL) containing 15% (w/v) PEG 4000, 8% (v/ v) tert-butyl alcohol, 0.3 M ammonium acetate, and 0.1 M sodium citrate at the same pH. Thin, platelike crystals appeared within 3 weeks. Using a single crystal, roomtemperature X-ray diffraction data were collected to 2.1 Å resolution on an ADSC Quantum 4R CCD detector at station PX14.1 of the Synchrotron Radiation Source (Daresbury, U.K.). All available data were indexed and integrated using
40-2.1 53460 8426 90.0 (89.7) 10.8 (58.4) 12.2 (2.0) 22.9-2.1 22.0 27.4 8425 1017 13 62 0.006 1.3 24.6 0.86 36.1 37.4 36.7 59.5 46.6
a Figures in parentheses correspond to the highest-resolution shell (2.18-2.10 Å). b Rmerge ) ∑h∑i|I(h) - Ii(h)|/∑h∑iIi(h), where I(h) and Ii(h) are the mean and ith measurements of the intensity of reflection h, respectively. c Rcryst ) ∑h|Fo - Fc|/∑hFo, where Fo and Fc are the observed and calculated structure factor amplitudes of reflection h, respectively. d Rfree is equal to Rcryst for a randomly selected 10% subset of reflections not used in the refinement (45).
DENZO (18), scaled using SCALEPACK (18), and converted to amplitudes using TRUNCATE (19). The crystal belonged to space group P21212 with the following unit cell dimensions: a ) 38.55 Å, b ) 64.20 Å, and c ) 61.25 Å. Detailed data processing statistics are presented in Table 1. Phase Determination and Refinement. Initial phases were determined by the molecular replacement method (20), employing as a search model the coordinates of Ang (PDB entry 1B1I) (21) modified by deletion of residues 58-70. Rotation and translation function searches were conducted using data in the range of 25-3 Å, and a clear solution was obtained which, after rigid-body refinement, had a correlation coefficient of 66.2% and an R-factor of 41.5%. CNS (22) was used to refine the model, adopting a maximum likelihood target for all procedures. Cycles of conjugate gradient minimization, simulated annealing, and geometry minimization were interspersed with calculation of sigmaA-weighted 2Fo - Fc and Fo - Fc electron density maps and manual model building using O (23). The resolution was increased stepwise, and B-factor refinement was carried out once all discernible protein residues had been incorporated. During the final stages of refinement, a citrate ion was introduced into the model and water molecules were added at positions where peaks exceeded 2.5σ in the Fo - Fc electron density map. Only those waters that were positioned at appropriate distances from suitable hydrogen-bonding partners and had temperature factors of 1 Å apart (ARH-I residues 1 and 2, 19 and 20, 87 and 88, and 90 and 91); these deviations are all far from the substituted region and the active site, and in most cases can probably be attributed to differences in the crystal packing arrangements of the two proteins. The 2Fo - Fc map of the guest segment in ARH-I is of high quality (Figure 2a), and reveals a topology strongly resembling that adopted by these residues in RNase A (Figure 1c). Alignment of the 15 CR atoms of this segment with the corresponding atoms of the 1.26 Å resolution structure of free RNase A (PDB entry 7RSA) (28) gives an rms deviation
Crystal Structure of Angiogenin-RNase A Chimera
Biochemistry, Vol. 41, No. 33, 2002 10485
Table 2: Potential Hydrogen Bondsa of the Guest Segment in ARH-I and Their Counterparts in RNase A
their RNase A counterparts especially closely. The one notable discrepancy between ARH-I and RNase A in this region involves Lys 65/66. The side chain of Lys 65 is not visible in ARH-I, indicating that it is highly mobile. Although Lys 66 has relatively weak density in some structures of free RNase A and RNase A-nucleotide complexes and is variable in orientation, its conformation is well-defined in each case (28-31). The Nζ atom of this residue lies within the active site region, usually near the main chain O of Asp 121, with which it hydrogen bonds in some instances (28, 29). Architecture of the B2 Site. Crystallographic, NMR, and mutational studies (3, 4, 29, 30, 32, 33) have shown that the B2 site of RNase A is formed by the side chains of Cys 65, Asn 67, Gln 69, Asn 71, Ala 109, Glu 111, and His 119. Interactions with the purine base include a functionally important hydrogen bond of Asn 71, transient hydrogen bonds of Asn 67, Gln 69, and Glu111, π-π stacking with the His 119 imidazole, and van der Waals contacts with Cys 65 and Ala 109. None of the residues from the segment of residues 65-72 of RNase A have structural analogues in Ang, and the Ang residue that corresponds to Glu 111 (Glu 108) is positioned 2-3 Å away from this region. Accordingly, Ang discriminates between bases at B2 much less effectively; e.g., its preference for A over G is ∼3-fold, compared with values of 12-25-fold for RNase A (1, 3). The base preference of the B2 site in ARH-I (29-fold in favor of A) strongly resembles that of RNase A (1). The ARH-I crystal structure now indicates that all of the various B2 components are likely to function similarly (Figure 3). Asn 70 superimposes well with Asn 71 in the RNase A‚ d(CpA) complex (PDB entry 1RPG) (29). The position of Asn 66 corresponds to one of two observed for Asn 67 in 1RPG (lower occupancy) and 7RSA (higher occupancy). The orientation of Gln 68 is different from that of its analogue Gln 69 after Cγ because of the proximity of a symmetryrelated Trp 91 residue, but Gln 68 should be free to reposition in solution; the conformation seen in ARH-I resembles that in the RNase A‚3′-CMP complex (PDB entry 1RPF) (29). Sγ of Cys 64 aligns well with the equivalent atom of RNase A. The other residues predicted to comprise the B2 site of ARH-I originate from the host portion. The Cβ atoms of ARH-I Ala 108 and RNase A Ala 109 superimpose well. The carboxylate of Glu 110 is almost 2 Å closer to B2 than is that of Glu 108 in Ang, and its O1 atom forms a watermediated hydrogen bond with Asn 70 Nδ2 which replicates an interaction in 7RSA. His 116 does not stack against a purine ring as does His 119, the equivalent residue, in RNase A complexes with substrate analogues (29, 30); instead, it adopts an alternative “B” conformation seen in some structures of free RNase A and RNase A complexes with pyrimidine nucleotides. In conformation B, the χ1 and χ2 torsion angles of the His side chain both differ by ∼160° from those for the productive “A” conformation (which is the conformation of His 114, the equivalent residue, in Ang). The orientation of His 116 in ARH-I appears to be dictated by crystal contacts with Trp 91 from a symmetry-related molecule, and in solution, this side chain would probably be able to play a constructive purine binding role as in RNase A. Impact of the Guest Segment on the Pyrimidine-Binding Site. In RNase A, residues 59-73 and the residues forming
RNase Ab
ARH-I bond
length (Å)
bond
Main Chain-Main Chain Interactions Ser 58 N-Ala 55 O 3.3 Ser 59 N-Ala 56 O Gln 59 N-Ile 56 O 3.0 Gln 60 N-Val 57 O c Lys 60 N-Ile 73 O Lys 61 N-Gln 74 O Val 62 N-Cys 71 O 2.6 Val 63 N-Cys 72 O Cys 64 N-Gln 68 O 2.9 Cys 65 N-Gln 69 O Gly 67 N-Cys 64 O 3.1 Gly 68 N-Cys 65 O Cys 71 N-Val 62 O 3.2 Cys 72 N-Val 63 O Tyr 72 N-Val 107 O 2.9 Tyr 73 N-Val 108 O Ile 73 N-Lys 60 O 3.2 Gln 74 N-Lys 61 O Val 107 N-Tyr 72 O 3.0 Val 108 N-Tyr 73 O Cys 109 N-Asn 70 O 3.1 Cys 110 N-Asn 71 O
length (Å) 3.0 2.9 3.3 3.0 2.8 2.9 3.4 2.8 2.8 3.0 3.0
Side Chain-Main Chain and Side Chain-Side Chain Interactions Ser 58 Oγ-Ala 55 O 3.0 Ser 59 Oγ-Ala 56 O -d Asn 61 Nδ2-Thr 69 O 3.1 Asn 62 Nδ2-Thr 70 O 2.8 Asn 61 Nδ2-Cys 71 O 2.7 Asn 62 Nδ2-Cys 72 O 2.9 δ2 δ2 Lys 65 N-Asp 118 O 3.2 Lys 66 N-Asp 121 O 2.8 Lys 65 Nζ-Asp 118 O Lys 66 Nζ-Asp 121 O -e 3.0 Asn 66 Nδ2-Gln 68 O1 -c Asn 67 Nδ2-Gln 69 O1 2.9 Gln 68 N-Asn 66 Oδ1 2.8 Gln 69 N-Asn 67 Oδ1 3.0 Gln 68 N2-Asn 70 Oδ1 -c Gln 69 N2-Asn 71 Oδ1 2.9 Asn 70 Nδ2-Cys 109 O 3.0 Asn 71 Nδ2-Cys 110 O 2.9 Lys 75 N-Gln 59 O1 2.5 Tyr 76 N-Gln 60 O1 2.9 a Hydrogen bonds are listed if the distance between a donor (D) and an acceptor (A) is shorter than 3.3 Å and the D-H-A angle is greater than 90°. b PDB entry 7RSA (28). c Distance of >3.3 Å. d Angle of 3.3 Å. d Angle of